hono-del/Africa-call-center

# CMC Call Center AI — Africa Distributor PoC (Phase 1) AI knowledge search & answer support mock for African automotive distributor call centers. This is a **demo mock**: all data is hard-coded sample data, and the "AI search" returns scripted responses (no real LLM calls). Design docs live in `docs/` (00–06). This build implements **Priority A** of `docs/06_implementation_plan.md`: - **Agent Assist** (`/assist`) — the core Phase 1 screen: intake → question input → AI search → answer candidates (multiple) / confirmation questions / references / next actions → outcome recording (required) → log saved - **Manager Dashboard** (`/dashboard`) — KPIs, channel/language/outcome splits, category breakdown, recent interactions (all computed from the sample logs) - **Demo sample data** — 40 inquiry logs weighted to reflect the real market (Toyota-heavy with Hilux on top, growing Suzuki, high grey-import ratio), including the **10 scripted demo cases** in `data/demoCases.ts` (engine no-start, warning lights, Bluetooth pairing, door lock, navigation language, service intervals, battery drain, TPMS, charging system, unknown service history) Out of scope in this build (Priority B/C): `/analysis`, `/suggestions`, i18n switching. The `/captions` screen from the original mock has been removed per the Phase 1 scope. ## Getting started Requirements: Node.js 18.17+ (or 20+). ```bash npm install npm run dev # open localhost (redirects to /dashboard) ``` Production build: ```bash npm run build npm run start ``` ## Demo walkthrough (≈3 minutes) Follows the flagship case in `docs/05_demo_scenario.md`: a customer in Côte d'Ivoire asks about parts lead time for a grey-import Toyota Hilux via WhatsApp. 1. **Open `/dashboard`** — show the KPI row (total interactions, resolution rate, AI answer adoption, satisfaction, knowledge gaps) and point out the grey-import share and the French/English split. 2. **Go to `/assist`** — the intake defaults are already set for the dem …